Vision Server DeepSeek Gemini Free Tier
FreeNot checkedVision Server DeepSeek Gemini Free Tier — Model Context Protocol server
About
Vision Server DeepSeek Gemini Free Tier — Model Context Protocol server
README
Adiciona analise de imagens ao Claude CLI usando modelos de texto puro (DeepSeek, etc.) via MCP + Gemini API gratuita.
Como funciona
Claude CLI (DeepSeek)
-> analyze_image("foto.png")
-> MCP Server (Python, stdio)
-> Gemini API (free tier, custo zero)
-> descricao textual
-> DeepSeek responde como se tivesse "visto" a imagem
Fallback automatico
O servidor tenta 4 combinacoes em cadeia antes de desistir:
key1 + gemini-3.1-flash-lite (RPM 15, RPD 500)
-> 429/503 -> key1 + gemini-2.5-flash-lite (RPM 10, RPD 20)
-> 429 cota -> key2 + gemini-3.1-flash-lite
-> 429/503 -> key2 + gemini-2.5-flash-lite
-> falha todas -> erro descritivo
| Erro | Significado | Acao |
|---|---|---|
| 403/401 | Chave invalida | Abandona a chave |
| 429 | Cota diaria esgotada | Proximo modelo/chave |
| 503 | Modelo saturado | Proximo modelo |
Instalacao
Pre-requisitos (todas as plataformas)
- Python 3.10+
- 1 ou 2 chaves API Gemini (gratuitas, sem cartao de credito)
- Claude CLI instalado
Obtenha as chaves em: https://aistudio.google.com/apikey
Instalacao automatica (recomendado)
Linux / macOS
chmod +x install.sh
./install.sh
O script pergunta as chaves API (1 obrigatoria, 1 opcional) e configura tudo automaticamente.
Windows (PowerShell)
Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass
.\install.ps1
O script pergunta as chaves API (1 obrigatoria, 1 opcional) e configura tudo automaticamente.
Instalacao manual
Caso prefira controle total, siga os passos abaixo.
Linux / macOS
python3 -m pip install mcp
mkdir -p ~/.claude/mcp-servers/vision-fallback
cp server.py ~/.claude/mcp-servers/vision-fallback/server.py
Adicione ao ~/.claude.json no objeto mcpServers:
"ai-vision": {
"type": "stdio",
"command": "python3",
"args": [
"/home/seu-usuario/.claude/mcp-servers/vision-fallback/server.py"
],
"env": {
"VISION_API_KEY_1": "sua-chave-1",
"VISION_API_KEY_2": "sua-chave-2"
}
}
Se tiver apenas 1 chave, remova a linha VISION_API_KEY_2.
Reinicie o Claude CLI.
Windows
python -m pip install mcp
mkdir %USERPROFILE%\.claude\mcp-servers\vision-fallback
copy server.py %USERPROFILE%\.claude\mcp-servers\vision-fallback\server.py
Adicione ao %USERPROFILE%\.claude.json no objeto mcpServers:
"ai-vision": {
"type": "stdio",
"command": "python",
"args": [
"C:\\Users\\SeuUsuario\\.claude\\mcp-servers\\vision-fallback\\server.py"
],
"env": {
"VISION_API_KEY_1": "sua-chave-1",
"VISION_API_KEY_2": "sua-chave-2"
}
}
Use o caminho absoluto real, nao %USERPROFILE%. Se tiver apenas 1 chave, remova VISION_API_KEY_2.
Reinicie o Claude CLI.
Verificacao
Apos reiniciar, abra o gerenciador de MCPs (/mcp no Claude CLI). O servidor ai-vision deve aparecer com 1 ferramenta (analyze_image).
Teste basico:
Descreva detalhadamente a imagem /caminho/para/foto.png
O resultado inclui metadados ao final:
Modelo usado: gemini-3.1-flash-lite
Cadeia tentada: key=AIzaSyAwR8kF... model=gemini-3.1-flash-lite
Modelos Gemini Free Tier (maio/2026)
| Modelo | RPM | TPM | RPD | Visao |
|---|---|---|---|---|
| Gemini 3.1 Flash-Lite | 15 | 250K | 500 | Sim |
| Gemini 2.5 Flash-Lite | 10 | 250K | 20 | Sim |
| Gemini 2.5 Flash | 5 | 250K | 20 | Sim |
| Gemini 3.5 Flash | 5 | 250K | 20 | Sim |
Consulte seus limites em: https://aistudio.google.com/rate-limit
Customizacao
Para usar modelos diferentes, edite as constantes no inicio do server.py:
MODEL_PRIMARY = "gemini-3.5-flash"
MODEL_FALLBACK = "gemini-2.5-flash"
Para usar apenas 1 chave API, deixe VISION_API_KEY_2 vazia. A cadeia pula entradas com chave vazia automaticamente.
Troubleshooting
| Problema | Solucao |
|---|---|
| MCP server nao aparece | Verifique python -c "from mcp.server import Server" |
| Erro 403 em todas as tentativas | Chave API invalida. Gere uma nova em aistudio.google.com |
| Erro 429 em todas as tentativas | Cota diaria esgotada para ambas as chaves. Aguarde reset (meia-noite UTC) |
| Erro "model is no longer available" | Atualize MODEL_PRIMARY/FALLBACK no server.py com modelos ativos |
Licenca
MIT — veja LICENSE.
from github.com/lupark3/MCP-Vision-Server-DeepSeek-Gemini-Free-Tier
Installing Vision Server DeepSeek Gemini Free Tier
This server has no published package — it is built from source. Open the repository and follow its README.
▸ github.com/lupark3/MCP-Vision-Server-DeepSeek-Gemini-Free-TierFAQ
Is Vision Server DeepSeek Gemini Free Tier MCP free?
Yes, Vision Server DeepSeek Gemini Free Tier MCP is free — one-click install via Unyly at no cost.
Does Vision Server DeepSeek Gemini Free Tier need an API key?
No, Vision Server DeepSeek Gemini Free Tier runs without API keys or environment variables.
Is Vision Server DeepSeek Gemini Free Tier hosted or self-hosted?
Self-hosted: the server runs locally on your machine via the install command above.
How do I install Vision Server DeepSeek Gemini Free Tier in Claude Desktop, Claude Code or Cursor?
Open Vision Server DeepSeek Gemini Free Tier on unyly.org, pick your client tab (Claude Desktop, Claude Code, Cursor) and press Install — the config is generated automatically, no JSON editing.
Related MCPs
GitHub
PRs, issues, code search, CI status
by GitHubFilesystem
Secure file operations with configurable access controls.
Memory
Knowledge graph-based persistent memory system.
Template MCP Server
A CLI tool to create a new Model Context Protocol server project with TypeScript support, dual transport options, and an extensible structure
by mcpdotdirectCompare Vision Server DeepSeek Gemini Free Tier with
Not sure what to pick?
Find your stack in 60 seconds
Author?
Embed badge for your README
Browse similar
All development MCPs
